Scientists have a lot of different tools for studying the brain. BrainTools is a series that aims to give you some background information on the different gadgets and methods that are being used by scientists today.
In a previous BrainTools post, I discussed magnetic resonance imagining (MRI), which is one of the most widely used tools for noninvasively (i.e. non-surgically) looking inside of our heads. While MRI is great for answering questions about the structure of the brain, it can’t say anything about what the different parts of the brain are doing, or whether different parts of the brain are important for different behaviors. To answer those kinds of questions, we need to use functional brain imaging.
You’ve probably seen pictures from functional imaging studies before, with brains all painted up with colorful blobs. At first glance, these pictures seem to be pretty easy to interpret: the colorful parts of the brain are involved in some specific behavior, and the parts of the brain that aren’t colorful aren’t involved. It’s great when science is so simple, right?
You can probably guess that interpreting functional imaging studies isn’t quite that straightforward. Our brains do not glow in bright colorful patterns when they’re working. We need to take a peek under the hood, and see exactly what functional imaging tools are measuring.
There are two kinds of functional brain imaging that are commonly used today: positron emission tomography (PET), and functional magnetic resonance imaging (fMRI). Both tools rely on the same underlying facts about the brain: first, oxygen is fuel for our brain cells; second, blood is the means by which oxygen is delivered to the brain. Based on this, there ought to be blood flowing to any part of the brain that is doing some kind of work. PET and fMRI both measure blood flow in the brain, but go about it in different ways.
In PET, a special radioactive chemical called a tracer is injected into the blood (at levels low enough to be considered safe). Tracers are specially designed so that they build up in specific parts of the body, like certain structures in the brain; they can even be tailored to attach to specific types of cells or proteins. Because tracers are radioactive, they decay over time, and emit gamma rays (radioactivity!) as they break down. PET scanners pick up these gamma rays, and can reconstruct the signals in three dimensions, leading to cool pictures like this (red areas show higher concentrations of tracer):
fMRI gets at the same problem as PET – measuring blood flowing in the brain – in a slightly different way. Remember, MRI uses a powerful magnet to construct images of the brain. Well, these powerful magnets can also be used to detect changes in blood flow, by capitalizing on the fact that our blood cells have a lot of iron in them, especially when they’re loaded up with oxygen. This gives us a measurement called BOLD, short for blood-oxygen-level dependent signal. fMRI pictures look like this (red blobs indicate stronger BOLD response):
In some respects, fMRI is a more sensitive tool than PET – it can detect changes that happen in smaller windows of time (a few seconds for fMRI, but many seconds or minutes for PET), and it can also detect changes in much smaller regions of the brain (a few millimeters for fMRI, but several centimeters for PET). On the other hand, the tracers used in PET studies can be tailored to selectively bind to things like neurotransmitters or tau proteins, which fMRI can’t do. Which tool you decide to use depends strongly on the questions you’re asking.
So, we’ve got tools that can measure blood flow in the brain – now we know exactly what parts are important for specific behaviors, right? Not so fast. You may have heard that we’re only using 10 percent of our brains at any given moment. Well, forget that vile lie right now; in reality, we are using all of our brain, all of the time (though some parts may be working harder than others). This means that there’s always blood flowing in every part of the brain, which makes it impossible to use blood flow alone to draw conclusions about which parts of the brain are involved in specific behaviors. But all hope is not lost!
At this point, we need to make another assumption: brain activity caused by specific behaviors is additive, known as the principle of pure insertion. For instance, let’s say that when you blink, your brain activity changes in a unique way, pattern A. Let’s also say that when you jump, you get a different pattern of brain activity, pattern B. Following the principle of pure insertion, we would expect the brain activity from a person who was simultaneously blinking and jumping, pattern C, to be the simple combination of the unique activity patterns from blinking and jumping, i.e. pattern A + pattern B.
Still with me? Now we get to do some arithmetic! If pattern A + pattern B = pattern C, then we can infer that pattern A = pattern C – pattern B, and that pattern B = pattern C – pattern A. This leads to another important principle in the design of functional imaging experiments, called cognitive subtraction. Since all of our brain is always active, we need to use cognitive subtraction if we want to isolate any specific brain patterns.
For example, let’s say you want to see which parts of the brain important for reading. To start, you might ask people to read short stories while scanning their brains. As a clever neuroscientist, you might also ask them to look at the same stories translated into foreign languages (languages that we know they can’t read). Both tasks involve looking at text on a screen, but only the first task involves reading. Following the principle of cognitive subtraction, if you then subtracted the brain activity elicited by task 2 from the brain activity in task 1, you’d be left with the pattern of brain activity representing the difference between the tasks, in this case, the brain regions involved in reading. Presto!
Pure insertion and cognitive subtraction are huge assumptions, and are still hotly debated by scientists today, but neuroimaging studies would be impossible to understand without them. Still, this is a good reason to be skeptical when reading about studies that “prove” some part of the brain is necessary to a particular behavior. To really understand the relationship between brain and behavior, we need converging evidence, that is, evidence from a lot of different kinds of experiments using different tools. One tool that can test the findings from functional imaging studies is noninvasive cortical stimulation, which I’ll address in a future installment of BrainTools.
Questions? Thoughts? Leave a note in the comments, or find me on Twitter, @jimkloet.